On statistical evaluation of the linearity in assay validation

Area of Science:

• Clinical Chemistry
• Analytical Chemistry
• Biostatistics

Background:

• Linearity is crucial for assay validation accuracy.
• Existing methods like CLSI EP6-A and Kroll's ADL have limitations in evaluating linearity.
• Current methods may suffer from inflated Type I error rates.

Purpose of the Study:

• To review and identify limitations in current statistical methods for linearity evaluation.
• To propose improved statistical procedures for assessing linearity in analytical methods.
• To ensure accurate and reliable linearity claims in assay validation.

Main Methods:

• Review of the CLSI EP6-A guideline and Kroll's ADL method.
• Formulation of linearity hypothesis as an alternative hypothesis.
• Development of a two one-sided test (TOST) procedure.
• Development of a corrected Kroll's procedure.
• Conducting simulation studies to compare methods.

Main Results:

• Current methods may have inflated Type I error rates.
• Proposed methods (TOST and corrected Kroll's) adequately control statistical size.
• Proposed methods demonstrate sufficient statistical power for linearity assessment.
• Simulation studies confirm the improved performance of the new procedures.

Conclusions:

• The proposed TOST and corrected Kroll's procedures offer a more statistically sound approach to linearity evaluation.
• These methods provide better control over error rates compared to existing techniques.
• Accurate linearity assessment is vital for reliable assay validation in laboratory settings.